Articles

The Thinking Teacher’s Guide to AI in the Classroom

By March 16, 2026May 7th, 2026No Comments

Published March 16, 2026 | Weekly AI in Education Newsletter

Last semester, one of my AP World History students submitted an essay about the causes of World War I that was technically flawless — structured, cited, grammatically clean. But something felt off. When I asked her to walk me through her argument in class, she stumbled. The ideas weren’t really hers. The AI had done the thinking.

That moment crystallized something I’ve been wrestling with ever since: AI in the classroom isn’t the problem. How we use it is.

We’re past the moment of deciding whether to use these tools. A recent EdWeek Research Center survey found that 61% of teachers used AI in some capacity in 2025, up from just 34% in 2023. The tools are here. Our students are using them whether we design for it or not. The question now is whether we’re going to be intentional about it — or just hope for the best.

What AI Actually Does Well in a History Classroom

I teach AP World History, which means I’m constantly asking students to wrestle with complex primary sources, competing historical interpretations, and the kind of ambiguity that makes history so rich — and so hard to teach. AI, used well, can actually make that harder work more accessible.

The team at Humy.ai has built a platform that lets students converse directly with historical figures — Churchill, Robespierre, Adam Smith. Students don’t just read about these people; they interrogate them, push back, and discover contradictions in their reasoning. It’s not a gimmick. When a student argues with a simulated Robespierre about the justification for the Terror, they’re doing real historical thinking.

Over at Virginia Tech, historians Allen and David Hicks have been working with in-service history teachers across the state on exactly this kind of AI integration. Allen put it plainly: “If we don’t learn more about how to use these tools, people who know less than we do about what quality education looks like are going to determine how they are used and implemented.” That’s the teacher’s version of “if you’re not at the table, you’re on the menu.”

The American Historical Association put it in writing last summer when their Ad Hoc Committee on AI in History Education approved a set of guiding principles. Their core message: historical thinking still matters, AI frequently “hallucinates” false information, and the solution isn’t to ban it — it’s to teach students to evaluate it critically. The AHA is right. And it maps perfectly onto what we already do in history class: interrogate sources, check for bias, ask who benefits from this narrative.

The Critical Thinking Paradox

Here’s the tension every teacher feels right now: 70% of teachers worry that AI weakens students’ critical thinking skills, according to a CDT report. And they’re not wrong to worry. Studies show that students who over-rely on AI can develop what researchers call “learned helplessness” — they stop pushing through hard problems because the tool will just do it for them.

But here’s the flip side: when AI gets something wrong, and it does, often, that’s actually a teaching moment. When a student asks ChatGPT about the causes of the Haitian Revolution and gets a confident-sounding but incomplete answer, they’re suddenly motivated to dig deeper. The AI’s overconfidence is pedagogically useful. It teaches discernment.

EdWeek covered this dynamic well in their October 2025 piece — and what struck me most was that many teens weren’t convinced the tools hurt their thinking. They saw themselves as capable of using them strategically. Maybe the answer isn’t to protect students from AI, but to teach them to use it the way a skilled researcher uses any imperfect tool: with skepticism, verification, and their own judgment in the driver’s seat.

Media Literacy Is the New Core Skill

Colleen Kenny made a case I keep coming back to in her February 2026 LinkedIn piece. She argues that generative AI has created “content chaos” — a collapse of shared interpretive frameworks. Trust in U.S. media has fallen to 28% (Gallup). The average person can no longer reliably distinguish real video from AI-generated video. When citizens can’t agree on basic facts, Kenny writes, “democracy itself becomes impossible.”

That’s not hyperbole. It’s a history teacher’s nightmare. We’ve spent decades helping students understand how propaganda worked in Nazi Germany or Cold War America. Now we’re handing them tools that can generate that propaganda in seconds, at scale, for free.

Kenny’s solution — and UNESCO’s, and the AHA’s — is the same: comprehensive media literacy, woven into every subject, not siloed into one unit or one class. History is actually the perfect place for this. We already teach source evaluation. We already ask: Who wrote this? When? Why? What’s missing? That’s media literacy. We just need to extend it to AI-generated content explicitly.

Where Teachers Are Gaining Ground

It’s not all anxiety. The CDT report also found that 69% of teachers say AI tools have improved their teaching methods, and 55% report having more time for direct student interaction because AI handles administrative tasks. That tracks with my own experience. Differentiated materials that used to take two hours now take twenty minutes. That’s two hours I can spend with students who actually need me.

The OECD’s Digital Education Outlook 2026 offers a useful caution here: AI used without pedagogical intention just enhances performance with no real learning gains. The teacher still has to design the experience. The tool doesn’t replace the thinking — it can create space for more of it, if we’re thoughtful.

That’s actually an opportunity. If AI handles the rote, repetitive stuff — summarizing background context, generating practice questions, drafting rubric language — then we can spend more classroom time on the things AI genuinely cannot do: facilitating debate, building community, helping students find meaning in the past.

The Pedagogical Spectrum: Where AI Fits

Not every AI use case belongs in every classroom. Here’s how I think about it across the spectrum:

Lower-order tasks (AI does the heavy lifting): Background research, vocabulary scaffolding, initial outlines, grammar checks on drafts. Fine — with transparency and citation.

Mid-level tasks (AI as thought partner): Generating counterarguments for debate prep, asking Socratic questions, providing feedback on essay structure before a teacher reads it. This is where AI really earns its keep in a thinking classroom.

Higher-order tasks (AI as interlocutor to interrogate): Evaluating AI-generated historical narratives for accuracy and bias. Fact-checking AI claims against primary sources. Arguing with a simulated historical figure and then critiquing the simulation itself. This is where the magic happens.

The AHA’s guiding principles call for transparent policies — being explicit with students about when AI use is encouraged, when it’s restricted, and always requiring citation when it’s used. That’s the right framework. Ambiguity breeds academic dishonesty; clarity breeds trust.

The Bottom Line

We are, as one commentator put it, past the “panic and pilot” years. 2026 is the year habits harden. The choices we make now — about how to integrate AI, what to protect, what to reimagine — will shape what our students are capable of for the rest of their lives.

My student who submitted that too-perfect essay? She’s one of my best writers now. Once we talked through what happened — and I made space to be honest about it without shaming her — she got interested in the challenge of using AI as a sparring partner rather than a ghostwriter. She started arguing back at it. That’s exactly what I want to see.

The tools are here. The question is whether we’re going to be the ones shaping how they’re used — or whether we’re going to step back and let someone else answer that.

I know what I choose.

Sources and further reading:

Share